tech
EU seeks to 'intensify' talks with U.S. on advanced cyber AI models, official tells CNBC, amid Mythos concerns
Anthropic's Mythos model, which has advanced cyber abilities, has prompted a wave of concern from governments and businesses.
TL;DR
- The EU aims to "intensify" discussions with the U.S. on advanced AI models, including those with cyber capabilities.
- Anthropic's "Mythos-class" models are expected to be released to customers soon, raising concerns about AI-powered cyberattacks.
- The White House is working with AI labs to balance innovation and safety, seeking to maintain its lead over China.
- Anthropic has not granted the EU preview access to Mythos, directing them to seek permission from the U.S. administration.
- U.S. officials reportedly opposed Anthropic's plan to expand access to Mythos due to security concerns.
- Cybersecurity is a shared priority, and the U.S. and EU have agreed to mutually recognize their standards in this area.
- Anthropic stated that strong cyber safeguards are required before general release of Mythos-class models.
- CEO Dario Amodei highlighted a time-sensitive window to address software vulnerabilities identified by Mythos before Chinese AI advances.